Incidence, Size and Spatial Structure of Clones in Second-Growth Stands of Coast Redwood, Sequoia Sempervirens (Cupressaceae)1

American Journal of Botany 91(7): 1140±1146. 2004.

INCIDENCE, SIZE AND SPATIAL STRUCTURE OF CLONES IN SECOND-GROWTH STANDS OF COAST REDWOOD, SEQUOIA SEMPERVIRENS (CUPRESSACEAE)1

VLADIMIR DOUHOVNIKOFF,2 ADELAIDE M. CHENG, AND RICHARD S. DODD

Department of Environmental Science, Policy and Management, University of California, Berkeley, California 94720-3112 USA

The ecology and evolutionary potential of coast redwood (Sequoia sempervirens) is signi®cantly in¯uenced by the important role clonal spread plays in its reproduction and site persistence. In nine second-growth stands, ampli®ed fragment length polymorphisms (AFLPs) were used to identify redwood clonal architecture. Clones (multistem genets) dominated sites by representing an average of 70% of stems measured, ranging in size from two to 20 stems. As a result, a relatively small number of genets can monopolize a disproportionate amount of site resources, are more likely to persist over time, and have greater on-site genetic representation. Clones were not limited to fairy-ring structures, but consisted of a wide range of shapes including concentric rings, ring chains, disjunct, and linear structures. Between-ramet distances of up to 40 m were measured, indicating that clonal reproduction is not limited to basal stump resprouting. Clonal structure in second-growth stands was similar to earlier reports from old growth, emphasizing the importance of site persistence and long-term, gradual site development. Smaller ramet numbers per genet in old growth is probably due to local within-genet self thinning. Management and conservation of redwoods will bene®t from a better understanding of the dynamics and structure of clonal spread in these forests.

Key words: AFLPs; clonal spread; genet; molecular markers; ramet; Sequoia; spatial structure; woody plant.

Asexual regeneration through sprouting from vegetative ma- son and Jerling, 1990). Evaluated ecologically, these bene®ts terial is a common phenomenon in plants (Stuefer et al., 2002), can result in high levels of productivity, in competitive ad- and many community types are dominated by such clonal spe- vantage for site colonization and persistence, and in an ability cies (van Groenendael and de Kroon, 1990; KlimesÏ et al., to survive a wide range of disturbances. 1997; Oborny and Kun, 2002). Commonly, clonal plants re- Through its in¯uence on plant ®tness, clonal spread also has produce both sexually, by the production and germination of important evolutionary implications. Clonal spread is the veg- seeds, and asexually, by sprouting from roots and shoots. De- etative establishment of genetically identical ramets that allow spite the fact that both means of reproduction play an impor- the persistence of a genet long after the original ortet is gone. tant role in the ecology and evolution of plant species (de There may be no limit to the number of ramet generations that Kroon and van Groenendael, 1990; Fischer and van Kleunen, can be generated over time within a clone, and each additional 2002; Pan and Price, 2002), the majority of research involving ramet adds to the overall ®tness of the genet through survi- plant population dynamics has focused on issues of sexual vorship, future seed production, and subsequent ramet produc- reproduction (fertilization, seed dormancy, longevity, morpho- tion (Fischer and van Kleunen, 2002; Pan and Price, 2002). logical characteristics, dispersal and germination requirements, In this way, clonal spread can in¯uence within- and among- etc.). This is re¯ected in the emphasis such reproduction is population genetic structure, effective population size, meta- given in most ecological and evolutionary theories and mod- population dynamics, natural selection, and possibly even the els. Yet, in many species the persistence of existing plant ge- evolution of geographic range (Eckert, 2002). notypes through clonal spread can be the more important av- Despite the potential importance of clonality and the fact enue to plant population regeneration and maintenance (Bond that many woody plants are capable of vegetative regeneration and Midgely, 2001). and clonal spread through sprouting (Peterson and Jones, The ecological signi®cance of clonal spread arises from the 1997), surprisingly few clonal studies have focused on trees bene®ts it confers to the plant. Some bene®ts include effective (Barsoum, 2002), and sprouting ``has received very little at- resource acquisition as a result of foraging (Oborny and Cain, tention in models and theories of forest succession or forest 1997), division of labor (Alpert and Stuefer, 1997), and re- diversity'' (Bond and Midgley, 2001, p. 49). Among woody source sharing (JoÂnsdoÂttir and Watson, 1997). Others include plants, the little work that has been reported has focused pri- better survival relative to seedlings of offspring in new and marily on angiosperms. sometimes dif®cult environments (Peterson and Jones, 1997), Redwoods (Sequoia sempervirens (D. Don) Endl.) are one reduced population turnover and dependence on seeds (Bond of the most productive timber types of the United States (Ol- and Midgley, 2001), and risk evasion on a genet scale (Eriks- iver et al., 1994) and provide an important supply of high- quality wood ®ber. Yet, beyond the fact that coast redwood is 1 Manuscript received 27 October 2003; revision accepted 16 March 2004. exceptional for a gymnosperm in its ability to readily resprout The authors thank Joe R. McBride and William L. Libby for their diverse and grow clonally after disturbances such as ®res, ¯oods, land- and important contributions, and William Baxter, Kevin L. O'Hara, Scott Ste- slides, and harvesting (Rydelius and Libby, 1993; Rogers, phens, John Battles, Matteo Garbelotto, and Nicholas Douhovnikoff for lab- oratory, ®eld, data, and manuscript assistance. This work was supported by a 1994, 2000), very little is known about the spatial structure of grant from the California Department of Forestry and Fire Protection. redwood clones, the relative importance of clonal vs. nonclon- 2 E-mail: [email protected]. al reproduction, or the variables that in¯uence clonal spread. 1140 July 2004] DOUHOVNIKOFF ET AL.ÐCLONAL SPREAD IN COAST REDWOODS 1141

Fig. 1. Sample sites of Sequoia sempervirens in the Jackson Demonstration State Forest, north coast of California, USA. Sites are represented by stars (river ϭ R1, R2, R3; north-facing ϭ N1, N2, N3; south-facing ϭ S1, S2, S3) and rivers are marked as dotted lines.

In a study of old-growth stands using allozyme markers, Rog- stands and explore the local structure of redwood clones. If ers (2000) found that a signi®cant number of redwood stems clonal spread is important in redwood stands, Bellingham and were not genetically distinct individuals, but were ramets of Sparrow (2000) predict that it should be in¯uenced by distur- larger genets, and thus resulted from cloning and not from bance frequency and site productivity. Therefore, we were also reproduction by seed. There are no similar studies of second- interested in determining if variability in broad site character- growth redwood forests despite the fact that in 1990 as much istics (i.e., aspect, disturbance levels) affects the importance as 96% of those lands in commercial redwood production of cloning and clone structure at a stand scale. available for harvest were made up of young or second-growth stands (California Department of Forestry and Fire Protection, MATERIALS AND METHODS 1990). If we are to properly understand the ecology and evolution- The study was conducted in the Jackson Demonstration State Forest (Fig. ary potential of redwoods, and clonal woody plants in general, 1) because of its combination of historic intensive management and the avail- it is essential to further explore the structure, dynamics, and ability of mature second-growth stands. A total of nine sites were selected (Table 1). Three sites each were selected on north-facing slopes (N), south- signi®cance of clonal spread in these forests. In the context of facing slopes (S), and on ¯oodplains adjacent to rivers (R). These broad cat- management and conservation, for example, this would result egories were chosen in order to investigate possible differences resulting from in more accurate estimates of minimum viable population size. variation in solar availability and moisture levels (north- vs. south-facing This is the ®rst major study of clonal spread in redwoods using slopes) and variation in disturbance regime (®re more common in the upland molecular genetic techniques. sites vs. ¯ooding in the riparian sites). Sites were widely distributed across It has been suggested that resprouters tend to be shorter than the forest, were at least 500 m from each other, and were pure stands of reseeders (Midgley, 1996); this is evidently not true when con- redwood estimated to have been last cut 50±70 yr ago. sidering coast redwoods. Our primary goals were to evaluate On each site, up to 45 adjacent trees over 10 cm in diameter at breast the importance of clonal spread in second-growth redwood height (dbh) were mapped, tagged, measured for dbh, and sampled for DNA.

TABLE 1. Site characteristics for Sequoia sempervirens stands in Jackson Demonstration State Forest, north coast of California, USA.

Area Stand Mean stem Surrounding sampled No. stems density diameter (cm) forest canopy Site (m2) sampled (no. stems/m2) (SE) character Site moisture Solar availability Riparian R1 1000 37 0.037 41.4 (3.8) Continuous Hydric±Mesic Moderate±Low R2 2600 45 0.017 41.6 (2.3) Continuous Hydric±Mesic Moderate±Low R3 1550 42 0.027 41.6 (4.3) Continuous Hydric±Mesic Moderate±Low North-facing N1 1600 42 0.026 28.7 (2.0) Continuous Mesic Low N2 4000 31 0.008 54.6 (6.6) Continuous Mesic Low N3 1500 43 0.029 40.6 (3.8) Continuous Mesic Low South-facing S1 1000 38 0.038 44.7 (6.4) Patchy Mesic Moderate±High S2 1700 37 0.022 48.0 (2.5) Patchy Mesic Moderate±High S3 1000 41 0.041 43.9 (3.3) Patchy Mesic Moderate±High 1142 AMERICAN JOURNAL OF BOTANY [Vol. 91

Sampling included clumped and forked stems that met the size requirement Clonal importanceÐTo assess the relative importance of clonal spread on at forking point. With a target of up to 45 trees sampled per site, different each site, sample percentage distinguishable (PD) values were calculated ac- stand densities resulted in a variation in total site areas (Table 1). In a few cording to Ellstrand and Roose (1987). PD values are calculated as the number instances we were unable to collect leaf material due to the extreme height of genets identi®ed divided by the total number of stems genetically sampled. or canopy structure of a tree. Mean stem diameters were measured for com- This metric of clonal diversity measures what percentage of all trees sampled parison in parallel plots of the same size outside the sampled stands and no is genetically distinct. signi®cant differences were found. The size of genets was measured in two ways. First, the total number of ramets per genet was tabulated to evaluate unit size. Second, aggregate basal areas per genet were calculated to serve as a surrogate for biomass. DNA extractionÐDNA was extracted according to the Cullings (1992) modi®cation of Doyle and Doyle (1987). DNA concentrations were estab- Clonal spatial structureÐTo test for spatial clustering of clones, we com- lished by electrophoresis on agarose gels and comparisons with lambda stan- pared the clonal structure based on our threshold approach with the spatial dards. pattern of samples at each of the ®eld sites by means of Mantel tests. For each of the ®eld sites, Mantel R correlations were computed between Euclid- AFLP analysisÐThe ampli®ed fragment length polymorphism (AFLP) ean distance matrices calculated from the point coordinates for all sampled method developed by Vos et al. (1995) was performed with the following individuals and binary matrices, in which pairs of individuals assigned to the modi®cations: Restriction digestion and ligation were performed simulta- same clone were given the value 1 and all non-clone pairs were given the neously in a 50 ␮L solution containing 250 ng of genomic DNA, 5 units (U) value 0. Signi®cance of the correlations was tested by 1000 random permu- of EcoRI,5UofMseI, 1 ϫ restriction-ligation buffer, 1 U T4 DNA ligase, tations. Mantel R correlation coef®cients were Z transformed to produce a 0.2 mmol/L ATP, 1.0 ␮mol/L MseI adapter, and 0.1 ␮mol/L EcoRI adapter. parameter with an approximate normal distribution (Sokal and Rohlf, 1995) The restriction-ligation reaction was incubated for 4 h at 37ЊC, then diluted and homogeneity among the Z-transformed Mantel R correlations for all site to 200 ␮L in Tris-EDTA buffer. Preampli®cation was performed in a 25 ␮L types was tested using a chi-square distribution. solution containing 2.5 ␮L of diluted restriction-ligation product, 0.2 mmol/ L dNTPs, 0.3 ␮mol/L of each primary ampli®cation primer, 1 ϫ polymerase RESULTS chain reaction (PCR) buffer, and 0.5 U Taq polymerase. For the primary ampli®cation primers, EcoRI primer was identical to the adapter sequence, The incidence and size of clonesÐA total of 356 redwood whereas the MseI primer had an extra ``C'' as a selective nucleotide. The PCR stems were genetically sampled over all nine sites. For each reaction was performed on a Techne Genius thermocycler for 28 cycles using stem, a genetic ®ngerprint was produced, allowing for its iden- the following cycling parameters: 30 s at 94ЊC, 60 s at 60ЊC, and 60 s at ti®cation and comparison to other stems. When single stem 72ЊC. The primary ampli®cation product was then diluted to 250 ␮L in Tris- genets (SSGs) were distinguished from multistem genets EDTA buffer. Selective ampli®cation was performed in a 25 ␮L solution (clones), a total of 35 clones (24% of all genets) and 113 SSGs containing 6.25 ␮L diluted primary ampli®cation product, 0.2 mmol/L dNTPs, were identi®ed on the nine sites sampled (Table 2). In general, 0.06 ␮mol/L EcoRI ¯uorescent selective primer, 0.3 ␮mol/L MseI selective clones were large, ranging in size from two to 20 ramets with primer, 1 ϫ PCR buffer, and 0.5 U Taq polymerase. We prescreened 32 a mean of 6.7 ramets per clone (SE ϭ 0.75). selective primer pairs and chose four pairs that were reproducible and ade- The PD values are an inverse measure of the importance of quately polymorphic (Ͼ100 polymorphic loci per primer) for this study (MseI- clonal spread on a site. The smaller the value, the greater is CCAC/EcoRI-ACG, MseI-CACA/EcoRI-AC, MseI-CCAA/EcoRI-CCC, and the proportion of stems that are replicate ramets of clones. In MseI-CCAC/EcoRI-CAC). The selective PCR reaction had two cycle sets: 13 this study, PD values were as low as 0.19, with an all-site Њ Њ cycles of 30 s at 94 C, 30 s at 65 C (annealing temperature was lowered mean of 0.43 (SE ϭ 0.05). On average, 32% of the stems on 0.7ЊC at each cycle), and 60 s at 72ЊC, followed by 18 cycles of 30 s at 94ЊC, Њ Њ a site were SSGs. The remaining 68% were ramets of an av- 30sat56C, and 60 s at 72 C. Fingerprint data were obtained by running erage of four clones per site. the ampli®ed samples on an ABI Prism 3100 DNA sequencing system (PE No signi®cant difference was detected in stem basal area Applied Biosystems, Foster City, California, USA), including a size standard ϭ 2 in each lane, using PE Applied Biosystems protocols. Band scoring was com- between individual clone ramets (mean 0.18 m ) and SSGs ϭ 2 ϭ ϭ pleted with the Genescan and Genotyper software (PE Applied Biosystems). (mean 0.20 m ) within sites (F ratio 0.36, P 0.55). However, on a whole genet basis, mean clone basal area per site, 1.36 m2 (SE ϭ 0.26), was more than six times greater Clone identi®cationÐBecause of the very large numbers of loci that can than mean SSG basal area per site (0.20 m2,SEϭ 0.03). The be detected, the AFLP method can be a very sensitive ®ngerprinting technique SSGs represented an average of 31% (SE ϭ 0.05) of total with enormous potential for assigning individuals to clones. However, choice basal area of all trees sampled (Fig. 2). Among multiple of loci is critical; too few or low levels of polymorphism may result in lack stemmed genets and SSGs, a full range of ramet diameters of resolution of genetically distinct, but closely related individuals, and too was found. This may have been a result of differential devel- many or excessive levels of polymorphism may reveal confusing within-genet opment due to resource availability and competition or to a differences. To minimize these risks, we selected loci that were polymorphic difference in ramet initiation time. within sets of full sibs and developed a statistical method for clone identi®- cation (Douhovnikoff and Dodd, 2003). Using this methodology, three control sets of four known full sibs and ®ve control sets with six clonal replicates Clone structureÐIn most cases, members of clones formed were used to calibrate a clone identi®cation threshold. Brie¯y, clonal pairwise partial circles surrounding large stumps, in a ``fairy-ring'' Jaccard similarities and full sib pair Jaccard similarities from AFLP ®nger- structure. However, in several cases not all stems in a fairy prints resulted in overlapping frequency distributions. The threshold similarity ring were identi®ed as part of the same clone (Fig. 3d). Also, (0.974) at, or above which, two sample ®ngerprints were considered to belong clones were not limited to a single ring structure. In fact, a to the same clone was determined by equating the lower tail of the clonal wide range of structures was detected, including ®gure eights distribution (minimum of 0.969) with the upper tail of the full sib distribution or chains of multiple overlapping rings, each centered on a (maximum of 0.977). The full sib and clonal replicates used to determine this different stump as found on site S2 (Fig. 3a); concentric rings threshold were 24-yr-old trees planted at the University of California Russell within larger rings as found on site S1 (Fig. 3b); disjunct Reservation Experimental Field Station, Lafayette, California. clones with no clear pattern, or even stumps from which they July 2004] DOUHOVNIKOFF ET AL.ÐCLONAL SPREAD IN COAST REDWOODS 1143

TABLE 2. Data for nine sites across three Sequoia sempervirens site types, in Jackson Demonstration State Forest.

No. SSGa Basal area SSG Site No. genets No. clones (percentage stem total) (percentage total) PD value Mantel Riparian R1 9 6 3 (8%) 14% 0.24 Ϫ0.41 R2 17 6 11 (24%) 30% 0.38 Ϫ0.42 R3 21 2 19 (45%) 34% 0.50 Ϫ0.11 North-facing N1 25 3 22 (52%) 44% 0.60 Ϫ0.14 N2 11 3 8 (25%) 30% 0.35 Ϫ0.45 N3 26 3 23 (53%) 45% 0.60 Ϫ0.34 South-facing S1 10 4 6 (16%) 20% 0.26 Ϫ0.63 S2 7 4 3 (8%) 9% 0.19 Ϫ0.59 S3 22 4 18 (54%) 57% 0.54 Ϫ0.22 Totals 148 35 113 (32%) 31% NA NA radiate as found on site R1 (Fig. 3c); and distantly disjunct ever, when all sites were considered independent of type, a clones with a small satellite cluster at distances as great as 40 positive correlation was found between PD values and Mantel m from the larger ring as found on site N2 (Fig. 3d). R correlations (R2 ϭ 0.674, P ϭ 0.006), indicating that as the Mantel correlations between spatial distances among stems importance of clonal spread on a site increased, within-clone on a site and assignment to clones ranged from Ϫ0.11 to clustering of ramets increased. In other words, each additional Ϫ0.63. All correlations were negative, indicating a tendency ramet tends be located near the other ramets in a genet. for stems of the same genet to be clustered. The chi-square tests revealed no signi®cant differences among site types in Variation in clone structure by site typeÐThe only signif- the Z-transformed Mantel R correlations, suggesting no sig- icant difference in clonal structure detected among site types ni®cant difference in spatial structure among site types. How- was maximum between-ramet distances, which were signi®-

Fig. 2. Percentage of total basal area represented by each clone and all single stem genets (singles) for each Sequoia sempervirens site (river ϭ R1, R2, R3; north-facing ϭ N1, N2, N3; south-facing ϭ S1, S2, S3) in the Jackson Demonstration State Forest. 1144 AMERICAN JOURNAL OF BOTANY [Vol. 91

of clonal events. On the other hand, it is possible to determine the minimum proportion of ramets that have resulted from clonal events. In this study, we found that on average 59% of the stems at a site resulted from clonal spread, with a high of 81% and a low of 40%. These are minimum values, and it is possible that 100% of the stems are clonal copies of predating stems that no longer exist. The skew in ramet number per genet is likely a result of the unevenness of disturbance events that open a window of opportunity for sprout release. For example, a disturbed tree (windfall, burned, cut, etc.) may clear an opening and promote sprouting at its base and possibly along its fallen stem result- ing in a large number of ramets in a particular clone relative Fig. 3. A selection of four genets that demonstrate the range of Sequoia to undisturbed genets. Regardless of the cause of this imbal- sempervirens clonal structures found in the Jackson Demonstration State For- ance it is most signi®cant that a few genets monopolize a site est, north coast of California, including (a) chains of overlapping rings on site as large clones. This has the following interesting ecological S2, (b) concentric rings on site S1, (c) disjunct stems on site R1, and (d) distantly disjunct stems on site N2. Solid circles represent ramets from a single and evolutionary implications: clone. Open circles represent trees not sampled, open diamonds represent single stem genets, crossed circles represent stumps, and dotted circles outline perceived fairy-rings. (1) Resource useÐSite resource use per genet is likely to be skewed toward larger clones. By virtue of sheer ramet num- bers, if all ramets draw on local resources equally, the genet cantly greater on the north-facing slopes (P ϭ 0.014). This with the most ramets gets most of the resources that are avail- was not due to a small-scale density difference, but appeared able. This is compounded by the fact that if the ramets in a to be due to a more common occurrence of disjunct ramets. clone remain integrated, resources can be shared across the Ramet clusters were found to be as far as 40 m from each genet. As a result, large redwood clones not only have a pro- other on these north-facing slopes. Maximum between ramet portionately larger footprint in terms of resource use, but they distances were positively correlated with the number of stems may also be taking advantage of economies of scale and great- in a clone (R2 ϭ 0.290, P ϭ 0.002) so that clones with larger er access to resources. Future work on the integration of red- numbers of ramets tended to have a greater overall footprint. wood clones and the extent and dynamics of resource sharing A weak but suggestive negative correlation was found be- is necessary to determine the larger scale physiological im- tween mean stem diameters at a site and Mantel R correlations portance of cloning in redwood. (R2 ϭ 0.389, P ϭ 0.072), implying that increased clustering of clonal ramets was associated with larger mean stem diam- (2) Site persistenceÐLarge redwood clones may have im- eters at the site. Accordingly, a weak negative correlation be- portant long-term advantages in maintaining site representa- tween site PD value and mean stem diameter at a site was also tion and dominance. This possibility is based on the following 2 ϭ ϭ detected (R 0.347, P 0.095), suggesting that as the im- reasoning: (A) Clones are genetic replicates of locally suc- portance of cloning at a site increased, mean stem diameters cessful genotypes giving them a competitive edge over new- at the site also increased. In summary, clonal importance, stem comers that are subject to the trial and error involved in sexual clustering, and stem diameters appear to be positively corre- reproduction. (B) A large proportion of redwood seeds are lated. sterile at the time of dissemination (Olsen et al., 1990). Those that are not have a very short period of viability (Fritz and DISCUSSION Rydelius, 1966), and the few seeds that do germinate are ex- Importance of clonal spread and clone structureÐOur tremely susceptible to damping off (Hepting, 1971). On the molecular analyses reveal that clonal spread is an important other hand, less seasonality is associated with sprouting, al- mode of regeneration in redwood stands. The PD values mea- lowing for a quick response to resource availability. Clones sured are similar to those of sandbar willows (Salix exigua have access to an already established root system, grow much Nutt.) (Douhovnikoff and Dodd, 2003) and many clonal non- faster than seedlings, and are not susceptible to damping off woody plant species such as bracken fern (Pteridium aquil- (Olsen et al., 1990). (C) In many cases, stochastic events act inum (L.) Kuhn) (Parks and Werth, 1993) and Vaccinium stam- upon clone ramets independently. As a result, the risk of mor- ineum L. (Kreher et al., 2000). The distribution of genet sizes tality to a genet can be spread among its many ramets (Cook, is skewed by several SSGs and a few disproportionately large 1983; Eriksson and Jerling, 1990). With more ramets, a red- clones with as many as 20 stems. As a result, 2±6 genets can wood genet is more likely to survive stem-removing distur- contain the great majority of stems on sites ranging in size bances. (D) Ramet clumping diversi®es stem location and can from 1000 m2 to 3000 m2. On these sites, the numbers of produce local characteristics that buffer some ramets from distinct genets were only about one-sixth of the total number damage (Peterson and Jones, 1997). For example, uphill red- of stems. The potential for high numbers of identical geno- wood ramets tend to be disproportionately damaged when fu- types is commonly overlooked in ecological studies, but our els that build up on the uphill side of clones burn in a ®re, results underline the need to take this mode of regeneration whereas the downhill stems are buffered from damage. (E) into account in clonal species such as redwood. Without a biological limit to the number of ramet generations We are unable to determine if a stem was the direct result over time that can be produced vegetatively, a successful red- of a seeding event. Even an SSG may be the last in a series wood genotype is locally persistent and potentially immortal. July 2004] DOUHOVNIKOFF ET AL.ÐCLONAL SPREAD IN COAST REDWOODS 1145

(3) Greater genetic representationÐLarge redwood genets a harvest appears to be a one-time clearing of ramets, leaving have greater genetic representation on a site with more stems, potential sprouting sites (stumps, root systems, etc.) that are pollen, and seedsÐall important factors driving evolutionary the result of site history that predates this event. As a result, potential. When regeneration by seed is possible, larger clones the long-term development of large-scale clonal structure may have a greater likelihood of being a genetic contributor to the be more important than the short-term effects of harvesting. cohort. In fact, a harvest may simply resemble another in a long series ``Although clonal plants generally sprout, only a small frac- of disturbances (®re, ¯ood, etc.) that the genets have survived tion of woody sprouters are clonal and capable of vegetative over time. spread'' (Bond and Midgley, 2001, p. 45). The important dis- The most signi®cant difference between second-growth red- tinction here is that sprouting on a local scale emphasizes stem wood forests and old growth forests was the greater number replacement, whereas clonal spread emphasizes lateral ramet of ramets per genet found in the former. Second growth stands multiplication in addition to stem replacement. Our results had a maximum of 20 ramets per clone and the site with the show that redwoods are an example of the latter. We identi®ed greatest mean number of ramets per genet (5.29, including many distinct clonal structures, demonstrating great variability single stem ramets), whereas the old growth stands had a max- in how redwood clones spread. The well-recognized fairy imum of six ramets per clone, and site mean values of less rings were a standard unit of clonal structure typical of the than two. Presumably, gradual self-thinning has occurred with- local sprouter type, but at a larger scale, clones showed a wide in the denser clumps of ramets, resulting in fewer ramets per range of shapes with large numbers of ramets that were un- genet surviving through time. In the absence of disturbance likely to thin to a single replacement stem over time. For ex- and new resource availability over time, there is a decrease in ample, ramets from the same clone were found to be up to 40 the number of ramets per genet on a local scale. This is similar m apart from each other and, without the limitations of our to dynamics in stands of Populus tremuloides Michx. through sample size, may prove to be even more widely dispersed. time (Peterson and Squiers, 1995). Little is known about how redwood clonal spread occurs In this study few signi®cant differences in clonal size or beyond local basal sprouting. While fairy rings, concentric cir- structure were found by site type. This may have been a result cles, and ®gure eights or longer chains can be explained by of limited clone sample size, with many sample stems being repeated basal sprouting of new ramets, it is still unclear how identi®ed as clonal ramets. Rogers (2000) also found only disjunct structures come about. Rogers (2000) suggests that ``modest'' differences between upland and lowland. To more this may be the result of trees falling, branches being buried accurately assess site type differences, a more extensive sam- by the impact, and ®nally resprouting at distances equivalent pling scheme would be necessary that can both identify the in some cases to the average height of these trees. Olsen et al. extent of clones on a local scale, but also cover a large area (1990) point out that redwoods are able to sprout anywhere and capture a more signi®cant number of clones. The fact that along the stem and that this could also result in the establish- redwood genets can be very large and that there are no good ment of distant sprouts along fallen trees. Another possibility surrogates for clone identi®cation makes it dif®cult to econo- is that ramets may be produced from disturbed existing root mize on the sampling intensity. systems (Weber, 1990; Lavertu et al., 1994). In our second- growth stands, more complex patterns of clonal structure may ConclusionsÐIn clonally spreading species many classical have resulted from site disturbance during harvesting and the assumptions of how stands should be managed for biological movement of soil and slash. An example of this may be the diversity, maintenance of viable population size, regeneration, disjunct and somewhat chaotic pattern found on site R1 (Fig. and site productivity need to be reevaluated. Our results in- 3c). This is an important area for further research. dicate that only about one in six stems on these second-growth We were unable, from our data, to determine if redwood sites are distinct genets. This amounts to approximately a 16% clonal spread is of a phalanx or guerilla type. While Mantel reduction in effective population size relative to stem census test results suggested a phalanx structure with well-clustered size. Although in large populations such a reduction to effec- ramets, mapping revealed instances of well-spread ramets tive population size is unlikely to have important consequenc- more typical of a guerilla structure. It is most likely that red- es on genetic diversity, as populations become more restricted, woods combine phalanx and guerilla types of clonal spread. the reduction of census size to effective population size be- This emphasizes the need to complement tests of association, comes increasingly important. In redwood this is to some ex- such as Mantel tests used here, with direct mapping of clonal tent offset by its hexaploid genome, but it will still be impor- structure. tant to have a better understanding of the effects of clonal spread on minimum population sizes needed to maintain ex- Old growth vs. second growth and site variabilityÐCom- isting genetic diversity. Our AFLP data do not allow us to parison of our results with clonal structure of old-growth make these estimates as allele frequencies for these dominant stands (Rogers, 1994, 2000) provides insight into how clonal markers cannot be accurately estimated for a system that is structure may change as stands develop through time. These not in Hardy-Weinberg equilibrium. As in the earlier study by comparisons need to be treated with some caution as isozymes Rogers (2000), we are unable to determine the extent of re- were used for the identi®cation of ramets in old-growth stands. production by seed. Single stem genets could be of seedling These markers are less ef®cient for this purpose than the origin or be derived from a preexisting genet. Abnormalities AFLPs used in our second-growth work (Smulders et al., in embryo development (Buchholz, 1939) and low levels of 2002). pollen production are believed to contribute to poor seed re- In general, the results from old-growth sites were very sim- production in redwood. However, in the long term, rare seed- ilar to those from second-growth sites, both having similar ling recruitment may be suf®cient to maintain levels of genetic structures, such as multiple genets in a fairy ring, disjunct diversity (Bond and Midgley, 2001). Understanding the rela- ramets, and complex clone con®gurations. In a clonal forest, tive proportions of seed and clonal reproduction will be crucial 1146 AMERICAN JOURNAL OF BOTANY [Vol. 91 to a better understanding of stand structure and evolution in and litter removal on the regeneration of Populus tremuloides. Journal redwood. of Vegetation Science 5: 561±568. MIDGLEY, J. J. 1996. Why the world's vegetation in not totally dominated by LITERATURE CITED resprouting plants; because resprouters are shorter than reseeders. Ecog- raphy 19: 92±95. ALPERT,P.,AND J. F. STUEFER. 1997. Division of labour in clonal plants. In OBORNY, B., AND M. L. CAIN. 1997. Models of spatial spread and foraging H. de Kroon and J. van Groenendael [eds.], The ecology and evolution in clonal plants. In H. de Kroon and J. van Groenendael [eds.], The of clonal plants, 137±154. Backhuys, Leiden, Netherlands. ecology and evolution of clonal plants, 155±184. Backhuys, Leiden, BARSOUM, N. 2002. Relative contributions of sexual and asexual regeneration Netherlands. strategies in Populus nigra and Salix alba during the ®rst years of es- OBORNY, B., AND A. KUN. 2002. Fragmentation of clones: how does it in- tablishment on a braided gravel bed river. Evolutionary Ecology 15: 255± ¯uence dispersal and competitive ability? Evolutionary Ecology 15: 319± 279. 346. BELLINGHAM,P.J.,AND A. D. SPARROW. 2000. Resprouting as a life history OLIVER, W. W., J. L. LINDQUIST, AND R. O. STROTHMANN. 1994. Young- strategy in woody plant communities. Oikos 89: 409±416. growth redwood stands respond well to various thinning intensities. BOND,W.J.,AND J. J. MIDGLEY. 2001. Ecology of sprouting in woody plants: Western Journal of Applied Forestry 9: 106±112. the persistence niche. Trends in Ecology and Evolution 16: 45±51. OLSEN, D. F., D. F. ROY, AND G. A. WALTERS. 1990. Sequoia sempervirens BUCHOLZ, J. T. 1939. The embryogeny of Sequoia sempervirens with a com- (D. Don) Endl. redwoods. In R. M. Burns and B. H. Honkala [eds.], parison of the Sequoias. American Journal of Botany 26: 248±257. Silvics of North America. U.S. Department of Agriculture, Agricultural CALIFORNIA DEPARTMENT OF FORESTRY AND FIRE PROTECTION. 1990. A Handbook 654: 541±551. review of redwood harvesting another lookÐ1990. California Depart- PAN,J.J.,AND J. S. PRICE. 2002. Fitness and evolution in clonal plants: the ment of Forestry and Fire Protection, Sacramento, California, USA. impact of clonal growth. Evolutionary Ecology 15: 583±600. COOK, R. E. 1983. Clonal plant populations. American Scientist 71: 244±253. PARKS,J.C.,AND C. R. WERTH. 1993. A study of spatial features of clones CULLINGS, K. 1992. Design and testing of a plant-speci®c PCR primer for in a population of bracken fern, Pteridium aquilinum (Dennstaedtiaceae). ecological and evolutionary studies. Molecular Ecology 1: 233±240. American Journal of Botany 80: 537±544. DE KROON, H., AND J. VAN GROENENDAEL. 1990. Regulation and function PETERSON,C.J.,AND R. H. JONES. 1997. Clonality in woody plants: a review of clonal growth in plants: an evaluation. In H. de Kroon and J. van and comparison with clonal herbs. In H. de Kroon and J. van Groenen- Groenendael [eds.], Clonal growth in plants, 177±1186. SPB Academic, dael [eds.], The ecology and evolution of clonal plants, 263±290. Back- Hague, Netherlands. huys, Leiden, Netherlands. DOUHOVNIKOFF,V.,AND R. S. DODD. 2003. Intra-clonal variation and a sim- PETERSON,C.J.,AND E. R. SQUIERS. 1995. An unexpected change in spatial ilarity threshold for identi®cation of clones: application to Salix exigua pattern across 10 years in an aspen white pine forest. Journal of Ecology using AFLP molecular markers. Theoretical and Applied Genetics 106: 85: 847±855. 1307±1315. ROGERS, D. L. 1994. Spatial patterns of allozyme variation and clonal struc- DOYLE,J.J.,AND J. L. DOYLE. 1987. A rapid DNA isolation procedure for ture in coast redwoods (Sequoia sempervirens). Ph.D. dissertation, Uni- small quantities of fresh leaf tissue. Phytochemical Bulletin 19: 11±15. versity of California, Berkeley, California, USA. ECKERT, C. G. 2002. The loss of sex in clonal plants. Evolutionary Ecology ROGERS, D. L. 2000. Genotypic diversity and clone size in old-growth pop- 15: 501±520. ulations of coast redwood (Sequoia sempervirens). Canadian Journal of ELLSTRAND,N.C.,AND M. L. ROOSE. 1987. Patterns of genotypic diversity Botany 78: 1408±1419. in clonal plant species. American Journal of Botany 74: 123±131. RYDELIUS,J.A.,AND W. L. LIBBY. 1993. Arguments for redwood clonal ERIKSSON, O., AND L. JERLING. 1990. Hierarchical selection and risk spread- forestry. In M. R. Ahuja and W. J. Libby [eds.], Clonal forestry II: con- ing in clonal plants. In H. de Kroon and J. van Groenendael [eds.], Clonal servation and application, 158±168. Springer, Berlin, Germany. growth in plants, 79±94. SPB Academic, Hague, Netherlands. SMULDERS,M.J.M.,J.VAN DER SCHOOT,B.IVENS,V.STROME,S.CASTI- FISCHER, M., AND M. VAN KLEUNEN. 2002. On the evolution of clonal plant GLIONE,F.GRASSI,J.BOVENSCHEN,B.C.VAN DAM, AND B. VOSMAN. life histories. Evolutionary Ecology 15: 565±582. 2002. Clonal propagation in black poplar (Poulus nigra). In B. C. van FRITZ, E., AND J. A. RYDELIUS. 1966. Redwood reforestation problems: an Dam and S. Bordacs [eds.], Genetic diversity in river populations of experimental approach to their solution. Foundation for American Re- European black poplar, 39±52. CsiszaÂr Nyomda, Budapest, Hungary. source Management, San Francisco, California, USA. SOKAL,R.R.,AND F. J. ROHLF. 1995. Biometry. W. H. Freeman, New York, HEPTING, G. H. 1971. Diseases of forest and shade trees of the United States. New York, USA. U.S. Department of Agriculture, Agricultural Handbook 386: 658. STUEFER, J. F., B. ERSCHAMBER,H.HUBER, AND J. I. SUZUKI. 2002. The JOÂ NSDOÂ TTIR,I.S.,AND M. A. WATSON. 1997. Extensive physiological in- ecology and evolutionary biology of clonal plants: an introduction to the tegration: an adaptive trait in resource-poor environments? In H. de proceedings of Clone-2000. Evolutionay Ecology 15: 223±230. Kroon and J. van Groenendael [eds.], The ecology and evolution of clon- WEBER, M. G. 1990. Response of immature aspen ecosystems to cutting and al plants, 109±136. Backhuys, Leiden, Netherlands. burning in relation to vernal leaf-¯ush. Forest Ecology and Management KLIMESÏ, L., J. KLIMESOVA,R.HENDRIKS, AND J. VAN GROENENDAEL. 1997. 31: 15±33. Clonal plant architecture: a comparative analysis of form and function. VAN GROENENDAEL, J., AND H. DE KROON. 1990. Preface. In J. van Gro- In H. de Kroon and J. van Groenendael [eds.], The ecology and evolution enendael and H. de Kroon [eds.], Clonal growth in plants: regulation and of clonal plants, 1±29. Backhuys, Leiden, Netherlands. function. SPB Academic Publishing, Hague, Netherlands. KREHER, S. A., S. A. FORE, AND B. S. COLLINS. 2000. Genetic variation VOS, P., R. HOGERS,M.BLEEKER,M.REIJANS,T.VAN DE LEE,M.HORNES, within and among patches of the clonal species, Vaccinium stamineum A. FRIJTERS,J.POT,J.PELEMAN,M.KUIPER, AND M. ZABEAU. 1995. L. Molecular Ecology 9: 1247±1252. AFLP: a new technique for DNA ®ngerprinting. Nucleic Acids Research LAVERTU, D., Y. MAUFFETTE, AND Y. B ERGERON. 1994. Effects of stand age 23: 4407±4414.